For many decades academic libraries have purchased monographs on a ‘‘just-in-case’’ basis, hoping to select the best books that their budget could afford and trusting that these would be the right decisions for their patrons, with lLL (inter-library loan) service as a slow and cumbersome back-up. Before the availability of Amazon and other online booksellers as well as e-books, ‘just in case’ collection development was the only option. Typically, librarians would use three methods of selecting books under this model: approval plans with library-oriented booksellers, involving highly refined subject profiles; manual selection by the librarians based on publisher catalogs, book reviews, librarian-oriented selection publications like Choice magazine and similar sources of information; and individual title recommendations made by their faculty and other patrons.

Studies on the success of these methods have tended to report estimates of 40–65% success, where ‘success’ is measured as a book circulating one or more times in its first 5–10 years of acquisition. In the 1970s a study at Pittsburgh University found that 40% of books purchased seven years earlier had never circulated.1 At Bucknell, 39% of books purchased from 1991–1994 had never circulated.2 A study at Cornell University found that of books in their collection published in 2001, only 35.5% had circulated at all by 20093 and that 55% of books published since 1990 had never been checked out. Spitzform reported in 2011 that, at the University of Vermont, ‘fully 40% of our books had not been checked out for years after they began sitting on our shelves.’4 In the same year Cheung, working in a Hong Kong academic library with an exceptionally high overall circulation rate (121,000 circulations per year from a collection of about 470,000), nevertheless found that for books 10–15 years old, about 36% had never circulated, with wide variation from 9% to 76% by LC (Library of Congress) classification.5 Downey reported in 2012 that at Kent State University ‘39% of all librarian-selected print books have never circulated.’6 At Wake Forest University, Cramer ran ‘a rather crude analysis’ and concluded in 2013 that roughly 50% of books have not circulated since local collection of circulation data began in 1991 and 36% of the remaining 50% had circulated only once.7 Some researchers such as Fry8 have recently questioned the interpretation of this literature, but publishers’ own discounting practices for ‘back-file’ books suggest that they see significantly lower demand for books only a couple of years after initial publication, validating librarians’ experiences.

Such criticism garnered little practical attention among selector librarians. Perhaps this is because it offered no alternatives. As described by classic psychological ‘cognitive dissonance,’ which in this case is the need to avoid accepting data that suggested the librarians were not doing a ‘good job,’ librarians in institutions not specifically reported on in the studies could tell themselves that those statistics did not apply to them, that of course they were much more successful than that.

However, the introduction of e-books brought with it a new option: DDA (demand-driven acquisition), also sometimes called PDA (patron-driven acquisition). A DDA program allows librarians to insert records for e-books that they have not actually purchased yet into their catalogs. When a patron clicks on a link to access one of the e-books, the purchase would be instantly triggered and the patron would have immediate access. DDA transforms ‘just-in-case’ purchases to ‘just-in-time’ purchases, with the patron unaware of the difference. Statistically, books purchased by DDA have a 100% use rate because the purchase is made at the moment of use.

A further option was quickly introduced, afforded by the nature of e-books: the concept of a use ‘rental,’ known in the industry as an STL (short-term loan). Libraries could opt to allow initial uses to be rented at a reduced cost (typically just 10–15% of the book purchase price) rather than trigger an immediate full purchase. Libraries could then select how many STLs would occur before a full purchase would be triggered. Thus some purchases could have not just a single use, but multiple uses, to justify the full purchase.

Case study – UPEI

The University of Prince Edward Island is a small (under 4,000 FTE) public comprehensive university in Canada, with a single library building and an annual acquisitions budget of approximately C$1 million, of which approximately C$140,000 is initially designated for monographs. The Library has six professional librarians who are assigned specific subject areas for selection responsibility, plus a University Librarian (the Library Director) who also participates in selection. Prior to the introduction of a DDA program, the Library did not use an approval plan, but books were selected individually by the librarians, who also accepted recommendations from faculty and other patrons. The monograph budget is allocated by LC classification rather than faculty department, which allows for more flexibility in dealing with interdisciplinary materials and requests.

The Library switched from a commercial ILS (integrated library system) in June 2008 to the open source Evergreen platform. This opens the data within the system to utmost analysis flexibility, as the holdings and circulation data can be queried with the full capabilities of the SQL query language.

Print circulation analysis

As of early 2016 only seven-and-a-half years (from June 2008) of circulation activity data is available from the ILS, as circulation data was not preserved during the migration. Fortunately, the migration did preserve the item creation dates (back to 1991). This is important because donated books, often acquired many years after they are published, can distort the data, and UPEI’s catalog does not preserve anonymous donation status, so they are indistinguishable from books selected by the librarians for purchase and thus mere publication dates are misleading as to circulation activity of purchased materials.

Table 1 provides circulation usage analysis for three different ranges of acquired book publication years. An analysis by individual LC class is also possible with the available data. (A detailed reporting of that is beyond the scope of this article, but there is substantial variation, ranging in the case of the 2008–2011 range from 34% success in the DSs to 73% in the SFs. [UPEI has a large veterinary program.])

Ratios of selected books to number checked out at least once at UPEI Library

Publication years included Item record creation date Total number of items added to collection Number that checked out 1+ times % of total checked out (rounded to nearest whole number)

2000–2003 2000–2003 5,725 1,927 33%
2005–2008 2005–2008 7,485 3,470 46%
2008–2011 2007–2013 7,184 3,612 50%

Table 1

Ratios of selected books to number checked out at least once. All circulation activity data is from the time period June 2008 – January 2016. Only books in the ‘stacks’ main circulating collection are included, not reference or reserves. UPEI does not collect in-house usage statistics.

Experience with DDA

In 2011, after first learning about the new DDA programs becoming available, the acquisitions librarian decided to experiment with the model. A modest amount of one-time money (C$3,000) had become available to support a new early childhood program in the Faculty of Education, one of the subject areas assigned to that librarian, so it was decided to use that to fund it. An account and profile were set up with YBP (Yankee Book Peddler), a ‘book jobber’ in the United States with extensive experience providing highly granular traditional approval plans that the company could now apply to the DDA model. The profile included an individual price cap of C$175 to prevent very expensive books, such as encyclopaedias, from entering the profile ‘pool.’ The account was configured to allow for two STLs before a purchase would be triggered. In the first year, of the C$3,000 deposited, and with 563 records loaded into the catalog, only C$345 had been spent. Almost all of that was spent on STLs, and only three books were actually purchased. This experiment allayed the other librarians’ fears that such a program could run away with the budget, and gradually more opted to add their assigned LC classes to the DDA profile, first in the sciences and social sciences, and eventually also in the humanities. The librarians did not consider carving out a separate budget for DDA purchases, as has been done at other institutions, but instead followed current UPEI practice of purchasing e-books from the same funds as print books, by LC class of the e-book itself.

In September 2012, when considering whether to renew the library’s subscription to a major academic e-book package, the librarians decided to try instead to spend that allocated money on an equivalently broad DDA program with a single e-book vendor (the same one licensing the package). It was also the platform for the library’s new discovery service, and the e-books would be discoverable and accessible through that search interface. The discovery service was highlighted on the Library’s home page as the best place to start most research for books and articles and would thus likely be searched far more heavily than the traditional catalog. The plan was for the vendor to send deposit account balance updates weekly. However, just five days after activating the new DDA offerings in the discovery service, the acquisitions librarian noticed that fully 50% of the funds in the deposit account were already spent. This was even before the cataloging unit had started to load the MARC records for the collection into the catalog, such was the exposure power of having the DDA in the discovery service. The librarians quickly agreed to halt the experiment, and find the extra money to restore the full amount needed to resubscribe to the package, thus affirming the cost effectiveness of such packages for a smaller library, and renewing appreciation for the highly specific control that the profile from YBP offered the DDA model.

In spring 2014 word was spreading through multiple channels that certain publishers had significantly increased the percentage of the purchase price that they would charge for STLs, causing the Library’s two STLs to cost as much as 50–60%, rather than 20–30%, of the purchase price in many cases (on top of the full 100% of the purchase price, as STL payments do not reduce the final retail purchase price). In response, the Library worked with YBP to identify those publishers in order to exclude them from the DDA program both retroactively and for future updates. It was hoped that if enough libraries did this, such price increases would result in less, not more, revenue, and the publishers would reconsider this practice.

By early 2015 approximately 11,000 DDA records had been loaded into the catalog. Over the entire 2011-2014 period, just C$4,275 had been spent, most of it on STLs, with just 33 books being fully purchased and an additional 240 books being used through STLs. The purchased books averaged C$69 per book (not counting the cost of their prior STLs), and the average cost of the STLs was C$7.26. It should be noted that since that time, many publishers have raised their STL price (by raising the percentage of the book purchase price that would be charged per STL).

As of early 2016 the Library now has all LC classes included in the profile and approximately 38,000 DDA records loaded in the catalog. This amounts to a theoretical retail value of well over C$2 million, but just C$4,200 has been spent on STLs (for 136 titles) and C$742 on purchases (for 11 titles) in the current fiscal year. The YBP program also enables the librarians to add further titles manually when the publisher allows them in DDA programs (some publishers do exclude some titles), but the uptake on this option has been anecdotally almost non-existent. (Data is not available to measure the number precisely.)

Further, after a statistical analysis was done in the summer of 2015, it was determined that, based on the past pattern of STL and purchase triggers, the Library would have saved money if there had been just one STL rather than two before full purchase. So the profile was adjusted to lower that trigger.

The Library has also just begun a ‘print DDA’ program. In this program, which has been running since fall 2015, librarians can individually select books that are not available as e-books and submit them to the cataloging unit, which would then create a special catalog record with a link to a request form that promises ‘rush’ handling. A workflow has been arranged between the ILL office and the acquisitions unit so that when a purchase is requested via this form, if ILL can get the book from the regional lending consortium faster than acquisitions can rush order it from Amazon, both events will happen: ILL will get the book for the patron even as acquisitions also purchases it. Otherwise, the acquisitions unit provides the book for the patron.

The print DDA program was considered an important addition because industry reports still indicate that upwards of 50% of scholarly books are not available as e-books within the first six months of their being released as print books. This program also took advantage of an existing workflow developed in 2014, called ‘ILL to purchase.’ The workflow involved the ILL office co-ordinating with the acquisitions unit to purchase rather than borrow any books requested by UPEI patrons that the ILL office could not get quickly and for free from the regional ILL consortium but that acquisitions could rush purchase faster and for a reasonable cost. In keeping with past budget practice of ILL having its own budget for paying borrowing costs charged by some lending libraries, a separate budget of about C$10,000 was initially set aside for the ILL-to-purchase program so that it would not impact the individual allocations of the LC-based funds. After the first year’s experience, the budget was reduced to C$3,000, and, with the end of this fiscal year fast approaching, just C$700 has been spent from it so far.

E-book collection philosophy and UPEI’s special constraints

As has been thoroughly debated throughout the literature of academic librarianship as well as popular media, e-books have advantages and disadvantages compared with print books for readers as well as libraries. In our experience at UPEI, the acceptance of e-books so far by both students and faculty is mixed, and not always along the discipline lines that stereotypes might predict. It is therefore the official collection development policy of UPEI’s library that the choice of format is one to be made by the subject librarians because there are such a large number of issues at play beyond the aesthetic response of any particular user.

In the summer of 2014 and again in summer 2015 this decision was made unassailable due to a series of outbreaks of mould in the Library’s compact storage facility and the subsequent determination that first, the space was no longer suitable for long-term print materials storage and second, no other space either on campus or off campus was available. This means that the Library is now faced with a massive weeding project. Approximately 80,000 volumes from the existing print collection of 500,000 will have to go, and all the librarians understand that in the face of having to discard such large numbers of existing books that have no electronic equivalent, the need to save precious shelf space by opting for electronic format over print whenever reasonably possible becomes paramount and must take precedence over personal preferences.

In addition to the shelf space issue, there have been significant budget shortfalls that a just-in-time purchase model can help address. Over the course of the last three years, Canada has seen its dollar drop 30% in value compared with the US dollar, which is the currency that almost all of its major materials, monograph and subscription, are priced in. For two consecutive years, the Library had to freeze most non-DDA monograph purchases until very late into the fiscal year when the final invoices of all major subscriptions were received and the librarians could see how much money would remain for monograph purchases. Because of the high ratio of selection to actual purchase experienced in its DDA programs, DDA ‘selection’ costs far less without reducing patron access than firm orders. Exceptions are made for a very small number of fields, such as veterinary medicine, in which UPEI is the library of record for the Atlantic Canada region, and for new DDA programs as described below.

Going forward

Unfortunately, as was much discussed at the 2015 ‘Charleston Conference’ by both DDA vendors and various e-book session presenters, it appears that the major scholarly publishers consider the entire STL ‘experiment’ to be a failure and will be moving to eliminate it completely as an option. As can be seen from the anecdotal data presented above, this will almost certainly be to the financial detriment of libraries like that of UPEI.

In the meantime, publishers are increasingly experimenting with EDA (evidence-based acquisition). Under this alternative model, the library commits to a minimum deposit ‘spend’ and then their patrons are given unlimited access to a significant portion of the publisher’s e-books (the amount of the commitment determining the portion of the e-books, often grouped by subject collection). Then, at the end of a fixed time period (usually either six months or one year), the librarians are provided with the usage data and are required to spend the entire deposit, but are free to make their own choices by title, using or ignoring the usage data as preferred. A new deposit is then made and the cycle starts again. Thus publishers are ensured a guaranteed source of revenue regardless of usage, unlike the STL/DDA model. The EDA model also so far excludes e-book aggregators like EBSCO and ProQuest from participating, as the e-books are purchased directly for the publisher’s own platform rather than for access on that of aggregators. This is not always the best choice for patrons, as the aggregator platforms typically offer much richer features and discovery integration with other databases and services. Some publishers also have such high demands for the ‘spend,’ or require such a small subset to be available for a more affordable deposit, that the patron-demand discovery aspect is all but lost.

Finally, the concept of DDA has now spread beyond monographs, with the recent model offered by the streaming video service, Kanopy. Kanopy blends DDA (complete with a ‘free three STL then purchase’ model) with the more typical (for video services) subscription model, such that a triggered ‘purchase’ is actually just a fee for one year’s access to that video, and if re-triggered in subsequent years would be invoiced again. UPEI is now participating in this service as well. It will be interesting to see if this model can translate back into the monograph market, as this concept of an annual subscription payment merged with DDA selection could have great value for more expensive materials such as large reference sets and scientific works.

Abbreviations and Acronyms

A list of the abbreviations and acronyms used in this and other Insights articles can be accessed here – click on the URL below and then select the ‘Abbreviations and Acronyms’ link at the top of the page it directs you to:

Competing interests

The author has declared no competing interests.